kernel.git/mm/swap.c, branch linux-5.7.y

mm: huge tmpfs: try to split_huge_page() when punching hole

2020-04-07T17:43:41Z

Yang Shi writes: Currently, when truncating a shmem file, if the range is partly in a THP (start or end is in the middle of THP), the pages actually will just get cleared rather than being freed, unless the range covers the whole THP. Even though all the subpages are truncated (randomly or sequentially), the THP may still be kept in page cache. This might be fine for some usecases which prefer preserving THP, but balloon inflation is handled in base page size. So when using shmem THP as memory backend, QEMU inflation actually doesn't work as expected since it doesn't free memory. But the inflation usecase really needs to get the memory freed. (Anonymous THP will also not get freed right away, but will be freed eventually when all subpages are unmapped: whereas shmem THP still stays in page cache.) Split THP right away when doing partial hole punch, and if split fails just clear the page so that read of the punched area will return zeroes. Hugh Dickins adds: Our earlier "team of pages" huge tmpfs implementation worked in the way that Yang Shi proposes; and we have been using this patch to continue to split the huge page when hole-punched or truncated, since converting over to the compound page implementation. Although huge tmpfs gives out huge pages when available, if the user specifically asks to truncate or punch a hole (perhaps to free memory, perhaps to reduce the memcg charge), then the filesystem should do so as best it can, splitting the huge page. That is not always possible: any additional reference to the huge page prevents split_huge_page() from succeeding, so the result can be flaky. But in practice it works successfully enough that we've not seen any problem from that. Add shmem_punch_compound() to encapsulate the decision of when a split is needed, and doing the split if so. Using this simplifies the flow in shmem_undo_range(); and the first (trylock) pass does not need to do any page clearing on failure, because the second pass will either succeed or do that clearing. Following the example of zero_user_segment() when clearing a partial page, add flush_dcache_page() and set_page_dirty() when clearing a hole - though I'm not certain that either is needed. But: split_huge_page() would be sure to fail if shmem_undo_range()'s pagevec holds further references to the huge page. The easiest way to fix that is for find_get_entries() to return early, as soon as it has put one compound head or tail into the pagevec. At first this felt like a hack; but on examination, this convention better suits all its callers - or will do, if the slight one-page-per-pagevec slowdown in shmem_unlock_mapping() and shmem_seek_hole_data() is transformed into a 512-page-per-pagevec speedup by checking for compound pages there. Signed-off-by: Hugh Dickins Signed-off-by: Andrew Morton Cc: Yang Shi Cc: Alexander Duyck Cc: "Michael S. Tsirkin" Cc: David Hildenbrand Cc: "Kirill A. Shutemov" Cc: Matthew Wilcox Cc: Andrea Arcangeli Link: http://lkml.kernel.org/r/alpine.LSU.2.11.2002261959020.10801@eggly.anvils Signed-off-by: Linus Torvalds

mm: code cleanup for MADV_FREE

2020-04-07T17:43:38Z

Some comments for MADV_FREE is revised and added to help people understand the MADV_FREE code, especially the page flag, PG_swapbacked. This makes page_is_file_cache() isn't consistent with its comments. So the function is renamed to page_is_file_lru() to make them consistent again. All these are put in one patch as one logical change. Suggested-by: David Hildenbrand Suggested-by: Johannes Weiner Suggested-by: David Rientjes Signed-off-by: "Huang, Ying" Signed-off-by: Andrew Morton Acked-by: Johannes Weiner Acked-by: David Rientjes Acked-by: Michal Hocko Acked-by: Pankaj Gupta Acked-by: Vlastimil Babka Cc: Dave Hansen Cc: Mel Gorman Cc: Minchan Kim Cc: Hugh Dickins Cc: Rik van Riel Link: http://lkml.kernel.org/r/20200317100342.2730705-1-ying.huang@intel.com Signed-off-by: Linus Torvalds

mm: swap: use smp_mb__after_atomic() to order LRU bit set

2020-04-02T16:35:28Z

Memory barrier is needed after setting LRU bit, but smp_mb() is too strong. Some architectures, i.e. x86, imply memory barrier with atomic operations, so replacing it with smp_mb__after_atomic() sounds better, which is nop on strong ordered machines, and full memory barriers on others. With this change the vm-scalability cases would perform better on x86, I saw total 6% improvement with this patch and previous inline fix. The test data (lru-file-readtwice throughput) against v5.6-rc4: mainline w/ inline fix w/ both (adding this) 150MB 154MB 159MB Fixes: 9c4e6b1a7027 ("mm, mlock, vmscan: no more skipping pagevecs") Signed-off-by: Yang Shi Signed-off-by: Andrew Morton Tested-by: Shakeel Butt Reviewed-by: Shakeel Butt Acked-by: Vlastimil Babka Acked-by: Johannes Weiner Cc: Matthew Wilcox (Oracle) Link: http://lkml.kernel.org/r/1584500541-46817-2-git-send-email-yang.shi@linux.alibaba.com Signed-off-by: Linus Torvalds

mm/swap.c: not necessary to export __pagevec_lru_add()

2020-04-02T16:35:27Z

__pagevec_lru_add() is only used in mm directory now. Remove the export symbol. Signed-off-by: Wei Yang Signed-off-by: Andrew Morton Reviewed-by: Andrew Morton Link: http://lkml.kernel.org/r/20200126011436.22979-1-richardw.yang@linux.intel.com Signed-off-by: Linus Torvalds

mm: devmap: refactor 1-based refcounting for ZONE_DEVICE pages

2020-01-31T18:30:37Z

An upcoming patch changes and complicates the refcounting and especially the "put page" aspects of it. In order to keep everything clean, refactor the devmap page release routines: * Rename put_devmap_managed_page() to page_is_devmap_managed(), and limit the functionality to "read only": return a bool, with no side effects. * Add a new routine, put_devmap_managed_page(), to handle decrementing the refcount for ZONE_DEVICE pages. * Change callers (just release_pages() and put_page()) to check page_is_devmap_managed() before calling the new put_devmap_managed_page() routine. This is a performance point: put_page() is a hot path, so we need to avoid non- inline function calls where possible. * Rename __put_devmap_managed_page() to free_devmap_managed_page(), and limit the functionality to unconditionally freeing a devmap page. This is originally based on a separate patch by Ira Weiny, which applied to an early version of the put_user_page() experiments. Since then, Jérôme Glisse suggested the refactoring described above. Link: http://lkml.kernel.org/r/20200107224558.2362728-5-jhubbard@nvidia.com Signed-off-by: Ira Weiny Signed-off-by: John Hubbard Suggested-by: Jérôme Glisse Reviewed-by: Dan Williams Reviewed-by: Jan Kara Cc: Christoph Hellwig Cc: Kirill A. Shutemov Cc: Alex Williamson Cc: Aneesh Kumar K.V Cc: Björn Töpel Cc: Daniel Vetter Cc: Hans Verkuil Cc: Jason Gunthorpe Cc: Jason Gunthorpe Cc: Jens Axboe Cc: Jonathan Corbet Cc: Leon Romanovsky Cc: Mauro Carvalho Chehab Cc: Mike Rapoport Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm/swap.c: piggyback lru_add_drain_all() calls

2019-12-01T14:29:19Z

This is a very slow operation. Right now POSIX_FADV_DONTNEED is the top user because it has to freeze page references when removing it from the cache. invalidate_bdev() calls it for the same reason. Both are triggered from userspace, so it's easy to generate a storm. mlock/mlockall no longer calls lru_add_drain_all - I've seen here serious slowdown on older kernels. There are some less obvious paths in memory migration/CMA/offlining which shouldn't call frequently. The worst case requires a non-trivial workload because lru_add_drain_all() skips cpus where vectors are empty. Something must constantly generate a flow of pages for each cpu. Also cpus must be busy to make scheduling per-cpu works slower. And the machine must be big enough (64+ cpus in our case). In our case that was a massive series of mlock calls in map-reduce while other tasks write logs (and generates flows of new pages in per-cpu vectors). Mlock calls were serialized by mutex and accumulated latency up to 10 seconds or more. The kernel does not call lru_add_drain_all on mlock paths since 4.15, but the same scenario could be triggered by fadvise(POSIX_FADV_DONTNEED) or any other remaining user. There is no reason to do the drain again if somebody else already drained all the per-cpu vectors while we waited for the lock. Piggyback on a drain starting and finishing while we wait for the lock: all pages pending at the time of our entry were drained from the vectors. Callers like POSIX_FADV_DONTNEED retry their operations once after draining per-cpu vectors when pages have unexpected references. Link: http://lkml.kernel.org/r/157019456205.3142.3369423180908482020.stgit@buzz Signed-off-by: Konstantin Khlebnikov Reviewed-by: Andrew Morton Cc: Michal Hocko Cc: Matthew Wilcox Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm/swap.c: trivial mark_page_accessed() cleanup

2019-12-01T14:29:18Z

This avoids duplicated PageReferenced() calls. No behavior change. Link: http://lkml.kernel.org/r/20191016225326.GB12497@wfg-t540p.sh.intel.com Signed-off-by: Fengguang Wu Reviewed-by: Andrew Morton Acked-by: Vlastimil Babka Cc: Dave Hansen Cc: Liu Jingqi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm: introduce MADV_COLD

2019-09-26T00:51:41Z

Patch series "Introduce MADV_COLD and MADV_PAGEOUT", v7. - Background The Android terminology used for forking a new process and starting an app from scratch is a cold start, while resuming an existing app is a hot start. While we continually try to improve the performance of cold starts, hot starts will always be significantly less power hungry as well as faster so we are trying to make hot start more likely than cold start. To increase hot start, Android userspace manages the order that apps should be killed in a process called ActivityManagerService. ActivityManagerService tracks every Android app or service that the user could be interacting with at any time and translates that into a ranked list for lmkd(low memory killer daemon). They are likely to be killed by lmkd if the system has to reclaim memory. In that sense they are similar to entries in any other cache. Those apps are kept alive for opportunistic performance improvements but those performance improvements will vary based on the memory requirements of individual workloads. - Problem Naturally, cached apps were dominant consumers of memory on the system. However, they were not significant consumers of swap even though they are good candidate for swap. Under investigation, swapping out only begins once the low zone watermark is hit and kswapd wakes up, but the overall allocation rate in the system might trip lmkd thresholds and cause a cached process to be killed(we measured performance swapping out vs. zapping the memory by killing a process. Unsurprisingly, zapping is 10x times faster even though we use zram which is much faster than real storage) so kill from lmkd will often satisfy the high zone watermark, resulting in very few pages actually being moved to swap. - Approach The approach we chose was to use a new interface to allow userspace to proactively reclaim entire processes by leveraging platform information. This allowed us to bypass the inaccuracy of the kernel’s LRUs for pages that are known to be cold from userspace and to avoid races with lmkd by reclaiming apps as soon as they entered the cached state. Additionally, it could provide many chances for platform to use much information to optimize memory efficiency. To achieve the goal, the patchset introduce two new options for madvise. One is MADV_COLD which will deactivate activated pages and the other is MADV_PAGEOUT which will reclaim private pages instantly. These new options complement MADV_DONTNEED and MADV_FREE by adding non-destructive ways to gain some free memory space. MADV_PAGEOUT is similar to MADV_DONTNEED in a way that it hints the kernel that memory region is not currently needed and should be reclaimed immediately; MADV_COLD is similar to MADV_FREE in a way that it hints the kernel that memory region is not currently needed and should be reclaimed when memory pressure rises. This patch (of 5): When a process expects no accesses to a certain memory range, it could give a hint to kernel that the pages can be reclaimed when memory pressure happens but data should be preserved for future use. This could reduce workingset eviction so it ends up increasing performance. This patch introduces the new MADV_COLD hint to madvise(2) syscall. MADV_COLD can be used by a process to mark a memory range as not expected to be used in the near future. The hint can help kernel in deciding which pages to evict early during memory pressure. It works for every LRU pages like MADV_[DONTNEED|FREE]. IOW, It moves active file page -> inactive file LRU active anon page -> inacdtive anon LRU Unlike MADV_FREE, it doesn't move active anonymous pages to inactive file LRU's head because MADV_COLD is a little bit different symantic. MADV_FREE means it's okay to discard when the memory pressure because the content of the page is *garbage* so freeing such pages is almost zero overhead since we don't need to swap out and access afterward causes just minor fault. Thus, it would make sense to put those freeable pages in inactive file LRU to compete other used-once pages. It makes sense for implmentaion point of view, too because it's not swapbacked memory any longer until it would be re-dirtied. Even, it could give a bonus to make them be reclaimed on swapless system. However, MADV_COLD doesn't mean garbage so reclaiming them requires swap-out/in in the end so it's bigger cost. Since we have designed VM LRU aging based on cost-model, anonymous cold pages would be better to position inactive anon's LRU list, not file LRU. Furthermore, it would help to avoid unnecessary scanning if system doesn't have a swap device. Let's start simpler way without adding complexity at this moment. However, keep in mind, too that it's a caveat that workloads with a lot of pages cache are likely to ignore MADV_COLD on anonymous memory because we rarely age anonymous LRU lists. * man-page material MADV_COLD (since Linux x.x) Pages in the specified regions will be treated as less-recently-accessed compared to pages in the system with similar access frequencies. In contrast to MADV_FREE, the contents of the region are preserved regardless of subsequent writes to pages. MADV_COLD cannot be applied to locked pages, Huge TLB pages, or VM_PFNMAP pages. [akpm@linux-foundation.org: resolve conflicts with hmm.git] Link: http://lkml.kernel.org/r/20190726023435.214162-2-minchan@kernel.org Signed-off-by: Minchan Kim Reported-by: kbuild test robot Acked-by: Michal Hocko Acked-by: Johannes Weiner Cc: James E.J. Bottomley Cc: Richard Henderson Cc: Ralf Baechle Cc: Chris Zankel Cc: Johannes Weiner Cc: Daniel Colascione Cc: Dave Hansen Cc: Hillf Danton Cc: Joel Fernandes (Google) Cc: Kirill A. Shutemov Cc: Oleksandr Natalenko Cc: Shakeel Butt Cc: Sonny Rao Cc: Suren Baghdasaryan Cc: Tim Murray Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm: move mem_cgroup_uncharge out of __page_cache_release()

2019-09-24T22:54:11Z

A later patch makes THP deferred split shrinker memcg aware, but it needs page->mem_cgroup information in THP destructor, which is called after mem_cgroup_uncharge() now. So move mem_cgroup_uncharge() from __page_cache_release() to compound page destructor, which is called by both THP and other compound pages except HugeTLB. And call it in __put_single_page() for single order page. Link: http://lkml.kernel.org/r/1565144277-36240-3-git-send-email-yang.shi@linux.alibaba.com Signed-off-by: Yang Shi Suggested-by: "Kirill A . Shutemov" Acked-by: Kirill A. Shutemov Reviewed-by: Kirill Tkhai Cc: Johannes Weiner Cc: Michal Hocko Cc: Hugh Dickins Cc: Shakeel Butt Cc: David Rientjes Cc: Qian Cai Cc: Vladimir Davydov Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

mm: replace list_move_tail() with add_page_to_lru_list_tail()

2019-09-24T22:54:08Z

This is a cleanup patch that replaces two historical uses of list_move_tail() with relatively recent add_page_to_lru_list_tail(). Link: http://lkml.kernel.org/r/20190716212436.7137-1-yuzhao@google.com Signed-off-by: Yu Zhao Cc: Vlastimil Babka Cc: Michal Hocko Cc: Jason Gunthorpe Cc: Dan Williams Cc: Matthew Wilcox Cc: Ira Weiny Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds